bart-paraphrase-finetuned-xsum-v2

This model is a fine-tuned version of eugenesiow/bart-paraphrase on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2329
• Rouge1: 100.0
• Rouge2: 100.0
• Rougel: 100.0
• Rougelsum: 100.0
• Gen Len: 9.2619

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
No log	1.0	21	1.2954	66.7012	60.8612	66.5163	66.4352	13.2857
No log	2.0	42	0.6866	86.8284	82.7835	86.7208	86.784	9.5238
No log	3.0	63	0.4652	95.1892	93.5619	95.2567	95.1657	10.3095
No log	4.0	84	0.4280	97.7463	97.1782	97.8708	97.718	9.5
No log	5.0	105	0.3712	99.6435	99.5767	99.6435	99.6435	9.3571
No log	6.0	126	0.4451	99.2695	98.9418	99.1883	99.3506	9.3095
No log	7.0	147	0.3169	99.246	99.0232	99.246	99.4048	9.619
No log	8.0	168	0.2942	100.0	100.0	100.0	100.0	9.4048
No log	9.0	189	0.3105	100.0	100.0	100.0	100.0	9.1667
No log	10.0	210	0.3035	100.0	100.0	100.0	100.0	9.2619
No log	11.0	231	0.2983	100.0	100.0	100.0	100.0	10.5714
No log	12.0	252	0.2497	100.0	100.0	100.0	100.0	9.4286
No log	13.0	273	0.2911	100.0	100.0	100.0	100.0	9.1667
No log	14.0	294	0.2619	100.0	100.0	100.0	100.0	9.2143
No log	15.0	315	0.2510	100.0	100.0	100.0	100.0	9.2381
No log	16.0	336	0.2647	100.0	100.0	100.0	100.0	9.9048
No log	17.0	357	0.2438	100.0	100.0	100.0	100.0	9.2143
No log	18.0	378	0.2324	100.0	100.0	100.0	100.0	9.3095
No log	19.0	399	0.2296	100.0	100.0	100.0	100.0	9.3095
No log	20.0	420	0.2329	100.0	100.0	100.0	100.0	9.2619

Framework versions

• Transformers 4.19.1
• Pytorch 1.11.0+cu113
• Datasets 2.2.1
• Tokenizers 0.12.1